Topic
8 replies Latest Post - ‏2010-09-21T05:57:40Z by lakshuraghav
jentek
jentek
29 Posts
ACCEPTED ANSWER

Pinned topic CICS-TX AKCS abends

‏2010-09-10T21:28:06Z |
Does anyone have experience resolving AKCS abends (timeout)? We were getting them quite infrequently, then suddenly a lot after a DB2 index change. We since reversed out the index changes, but the abends continue!

Thanks

Steve
Updated on 2010-09-21T05:57:40Z at 2010-09-21T05:57:40Z by lakshuraghav
  • Vivekster
    Vivekster
    3 Posts
    ACCEPTED ANSWER

    Re: CICS-TX AKCS abends

    ‏2010-09-13T06:39:42Z  in response to jentek
    jentek,

    AKCS abend is a deadlock timeout abend that is specified in seconds. The deadlock may resolve itself, so try increasing the DeadLockTimeout attribute ( in the TD Stanza ) to the maximum possible value.
    Can you also attach the console and symrecs file of that region?

    Cheers,

    Vivekster
  • lakshuraghav
    lakshuraghav
    7 Posts
    ACCEPTED ANSWER

    Re: CICS-TX AKCS abends

    ‏2010-09-13T07:14:25Z  in response to jentek
    Hi Steve,

    I see below explanation about AKCS abend in the documentation. Have you checked the user response already? Have you checked whether any DB2 specific errors are recorded in db2 logs? We sometimes see the issue when a DB2 SQL query takes a long time to respond.Can you check if you are seeing any transaction spending long time on the database side.

    Explanation:
    CICS detected a deadlock timeout condition. This condition may occur within any transaction defined with a non-zero DeadLockTimeout attribute in the Transaction Definitions (TD). This condition can also be raised by DB2 when it detects a deadlock or timeout on its resources. Deadlock timeout occurs when a transaction waits for longer than the time specified in DeadLockTimeout. (Note: the DeadLockTimeout attribute is specified in seconds.)
    System action:
    CICS abnormally terminates the transaction but does not produce a transaction dump.
    User response:
    You should re-run the transaction because the situation causing the transaction to wait, may well clear itself. If this does not solve the problem, try increasing the value of the DeadlockTimeout attribute to the maximum acceptable value. If this does not solve the problem, please read Problem Determination. If SQLCODE -913 is reported in the symrecs file, refer to the relevant DB2 Problem Determination Guide for further assistance.

    Thanks
    Raghu
    • jentek
      jentek
      29 Posts
      ACCEPTED ANSWER

      Re: CICS-TX AKCS abends

      ‏2010-09-14T21:07:59Z  in response to lakshuraghav
      Hi Raghu

      We restarted DB2 to reset the deadlock and timeout counters. During the CICS sessions we saw AKCS abends BUT no increase in the DB2 counters, therefore can we assume it is NOT a DB2 issue?

      We do have some long SQL transactions, but we had those before we got a big increase in AKCS abends.

      Thanks

      Steve
  • SystemAdmin
    SystemAdmin
    308 Posts
    ACCEPTED ANSWER

    Re: CICS-TX AKCS abends

    ‏2010-09-13T07:37:34Z  in response to jentek
    >> We were getting them quite infrequently, then suddenly a lot after a DB2 index change.

    Can you elaborate on this change ? Are you using DB2 as filemanager ?

    regards,
    msk
    • jentek
      jentek
      29 Posts
      ACCEPTED ANSWER

      Re: CICS-TX AKCS abends

      ‏2010-09-14T20:59:20Z  in response to SystemAdmin
      No, we use SFS, DB2 is our user database

      Steve
  • lakshuraghav
    lakshuraghav
    7 Posts
    ACCEPTED ANSWER

    Re: CICS-TX AKCS abends

    ‏2010-09-21T05:57:40Z  in response to jentek
    Hi Steve,

    AKCS is very generic abend.AKCS abends are common when DeadlockTimeout is set and multiple transactions contend for a same resource.In this case, I suspect the DB2 to be the issue as the problem started appearing in your case after changing the indices of DB2 table. But AKCS can happen irrespective any kind of resource contention such as RM,TDQ,TSQ,ENQ etc. Probably , you can quickly check by enabling Monitoring facility in TXSeries to check if there are any waits on the cics side which is resulting in AKCS abend. With the Monitoring facility, you can identify if the problem is due to CICS resource or the DB2/external resource.

    Thanks,
    Raghu